What is a color space?
makingsoftware.com
makingsoftware.com
If I had to criticize, I wish it had talked a bit more about printer color profiles (although in this paperless, web-world we live in, perhaps printing is in fact esoteric).
Unlike displays, a printer can't be simply defined with three primaries and a white point. Printer profiles can be quite large as they rely on someone having printed out a copious number of swatches on a given paper type and then measured (with some kind of colorimeter) device independent color values for each swatch. Those are used to build a large table for mapping from device independent color spaces to the printer's gamut.
Those large tables make the profile so large. And then of course interpolation is still required when mapping from a device independent color space to the printer profile. (Now imagine too that you need a different profile for each type of paper you might want to print to since each can represent color differently — plain paper unable to get the levels of saturation that a coated paper can.)
What was shocking to me was just how small the gamut of a printer typically is when seen alongside that of a decent display.
Consider that, in print, you'll never see an image as vivid as you can display on a nice, modern display. (And then consider that there are colors in nature so vivid that even a modern display cannot represent them. Just look at how much color is outside the triangle on the CIE "shark fin" color representation.)
Also not touched on (did I miss it?), all the math presented to map from one color space to another, also allows for "soft proofing" — where in fact you might match to a printer ICC profile but then take the result and match again to the user's display to give you a "preview" of what will be lost when going to said printer.
What was a nightmare for me when I worked in prepress was how hard it is to get a convincing purple out of a $20K printer. I used tricks which basically produced nothing like purple in a way that gave customers a good purple impression because the things with a closer actual resemblance to purple always looked awful.
My father used to work on all sorts of R&D involving things like how much K to use in substitution of CMY without getting desaturated etc. It's a real rabbit hole, especially if you want to reduce the amount of ink used to prevent soaking the paper.
Check out "Basic Color Terms: Their Universality and Evolution" by Brent Berlin and Paul Kay.
It affects communication. People can still discern the difference between colors, they just don't have an easy way to communicate this difference to others.
The Japanese language until relatively recently didn't have a clear verbal distinction between what we call green and blue in English. That doesn't mean Japanese people can't tell the difference between green and blue. It just means that there is a kind of "blue" that is the sky and a kind of "blue" that is for traffic control lights, and in context nobody is confused.
The same issue can occur within a language between people with differing levels of study of color. A graphic designer might say a particular shade of green is "chartreuse" that his boss instead might call "yellowish green".
Bad example. It seems that there is still a lot of disagreement on this matter, much of it rooted in what constitutes a word.
A well-balanced article on the subject here:
https://linguisticdiscovery.com/posts/inuit-words-for-snow/
Edit…
Also, from that article, a fascinating study: ‘Russian blues reveal effects of language on color discrimination’
It is also true that among mammals, chromatic vision is pretty much restricted to primates. The ability to perceive difference of light is a must if you don't want to become someone else's food. In contrast, chromatic vision is an 'extra' that in many ways serves as a (literally) florid extension to our lives. To me it is no surprise that range of emotion and range of hue are so often associated with each other. Interestingly enough, they are similarly mapped: as a set of differences, rather than as a degree of intensity.
As for the "Russian blue" study, I find it strange that the article is so skeptical of unreplicated results in linguistics and yet seems to accept the "Russian blue" study uncritically. I can see at least one glaring flaw: all of the Russian speakers were bi-lingual with English, with at least some of them being so since they were young children. They also discarded 16% of their test data because they deemed responses "too slow", with this discarding more heavily weighted towards the Russian speakers.
You'd be surprised. It may be heading that way, to arrive in a couple more decades, but not yet. It's still quite the industry (that I'm in).
Maybe not appropriate for that particular article but definitely appropriate for the site
Even native still has tons of issues, like the fact that AFAICT, no OS does HDR screen capture. You're viewing an HDR image, you ask the OS to capture the screen. It gives you an SDR capture :( On Mac and iOS that's certainly true. On Windows, the XBox Game Bar thingy will actually capture HDR but the OS level PrintScreen method will not, and the popular ShareX will not either.
This affects brightness and contrast: For emissive displays, you can have emissive values that are several to many orders of magnitude brighter than the 'black point', and more importantly, the primaries are defined by the display, not by ambient illumination.
Part of the magic of HDR displays is manipulating local masking (a human perceptual quirk) to drive bright regions on a display much brighter than the darker regions, so you can achieve even higher contrast ratios than the base technology could achieve (LED back-illuminated LCD panels, for many consumer TVs). Basically, a bright pixel will cause other nearby pixels to be brighter, because you can't see the dark details near a bright region anyway — but other regions could be darker, where you can perceive more detail in the blacks. This is achieved by illuminating sections of the display at significantly higher or lower levels, based on what your eyes/brain can actually perceive. That leads to significantly higher contrast ratios.
(As a heuristic: photographers generally say you can only get ~5 stops of contrast out of a print. (That is, bright areas are 2^5 times brighter than the darkest regions.) Modern HDR displays can do 2^10 or better. YMMV.)
But this also affects color... much of the complexity in getting printers to match derives from the interaction between the imperfect gamut caused by differing primaries, as filtered through human perception (and/or perceptual models). But you can't control the ambient illumination, so you're at the mercy of whatever the spectrum of your illumination is, plus whatever adaptation the viewer has. This feels fundamentally impossible to do "correctly" under all circumstances.
Which is to say, the original sin of color theory is the dimensional collapse from a continuous spectrum to a 3-dimensional, discretized representation. It's a miracle we can see color at all...!
In itself that is correct, but as you've noted, our own vision system isn't operating like that. The same display brightness and colors will be perceived very differently depending on the ambient light's brightness and color, and can also mean a severe breakdown in the dynamic range that can be made visible via a display.
And this ambient light also clearly impacts how prints are seen.
Just started playing with it with my spectrometer based on one of the examples they have, to convert spectral data to a single RGB value.
There is one axis Grayscale - that gives you the Brightess value. You can use that for black/white TV.
Now for colour you have 2 dimensions left. For these you pick one axis, where the eye is most sensitive - and perpendicular to that you have one axis, where the eye is least sensitive.
This colour space is the oldest and all axis there make sense. You can easily compare brightness, and you e.g. can assign more bandwidth to the sensitive colour axis if you want.
I spent the majority of my career at an optical equipment manufacturer, and we wrote a color management system to handle 48-bit color, before any OS manufacturers had it.
Non-trivial stuff, but powerful.
I know of at least one technology that works by converting to an esoteric color space, messing with image data there, then converting back.
> So why do we have so many different color spaces?
I think there was a missing piece here. Different representations of colour are useful for different things. I'm not going to give any secrets away but...if your trade involves finding out how close one colour is to another, then something that represents colours as points in space could make the maths easier. Then if you wanted to know if one colour was brighter than another, then something that represents a colour with a separate 'brightness factor' would make that trivial.
https://austingwalters.com/chromatags/
Think of it this way, a QR code is binary. If you modify color spaces correctly you can get 6 bits (or more) per pixel. In addition, you can improve the detection at distance, localization for robots, and speed (120 fps).
Done this to great effect previously and you can do a lot of awesome things with it. Pretty much the easiest hack in computer vision.
But can someone explain this to me?
>Light is technically something called electromagnetic radiation and it has a frequency and wavelength. That wavelength can vary, depending on the energy of the wave. High energy waves have a higher frequency and shorter wavelength, and low energy waves have a lower frequency and longer wavelength.
>This means that the same amount of energy at different wavelengths will not be perceived as the same brightness. For example, a light with a wavelength of 555 nm (green) will appear brighter than a light with a wavelength of 450 nm (blue) even if they have the same energy.
The article asserts that the wavelength (thus color) changes with the amount of energy, but then it says that you can have light of different wavelengths (color) with "the same energy."
So a "blue wave" has more energy than a "red wave" if both have the same amplitude (blue has a shorter wavelength, and energy is inversely proportional to wavelength). But you can have a "blue wave" with the same energy as a "red wave" if you increase the amplitude of the "red wave" to compensate for its longer wavelength.
So that means blue has more energy because it pulsates faster, and in spite of this we're less sensitive to it than we are to red, which pulsates slower. It's like our light sensitivity forms some sort of bell curve.
These energies are very small. You can add a lot of photons per second, increasing the brightness of that color, and this now has the energy per second of all those photons. So you can have a lot of red photons which sum to some energy, or a different number of blue photons that sums to (very, very close) the total red energy.
These are the two energies he confuses in those two places.
A related concept is sound vs psychoacoustics[1]. Sound is just pressure waves, but what we hear is a perception and has all sorts of aspects like masking[2]. The pressure waves contains two different signals but thanks to masking we might only perceive one.
Personally I think color constancy[3] really drives home that color is a perception and not something fundamental like the wavelength of photons.
[1]: https://en.wikipedia.org/wiki/Psychoacoustics
On the other hand, we perceive temperature yet its impact on the physical world is universal.
They are as good a demonstration as any of the notional nature of color spaces.